Multivariate ENSO Index (MEI)

NOTE:  The MEI Update for Dec 2016 is Delayed Due to Issues with the Source Data


The views expressed are those of the author and do not necessarily represent those of NOAA.

Click to enlarge

Outline for MEI webpage (updated on November 8th, 2016)

This webpage consists of seven main parts, three of which are updated every month:

1. A short description of the Multivariate ENSO Index (MEI);

2. Historic La Niña events since 1950;

3. Historic El Niño events since 1950;

4. UPDATED MEI loading maps for the latest season;

5. UPDATED MEI anomaly maps for the latest season;

6. UPDATED Discussion of recent conditions;

7. Publications and MEI data access.

El Niño/Southern Oscillation (ENSO) is the most important coupled ocean-atmosphere phenomenon to cause global climate variability on interannual time scales. Here we attempt to monitor ENSO by basing the Multivariate ENSO Index (MEI) on the six main observed variables over the tropical Pacific. These six variables are: sea-level pressure (P), zonal (U) and meridional (V) components of the surface wind, sea surface temperature (S), surface air temperature (A), and total cloudiness fraction of the sky (C). These observations have been collected and published in ICOADS for many years. The MEI is computed separately for each of twelve sliding bi-monthly seasons (Dec/Jan, Jan/Feb,..., Nov/Dec). After spatially filtering the individual fields into clusters (Wolter, 1987), the MEI is calculated as the first unrotated Principal Component (PC) of all six observed fields combined. This is accomplished by normalizing the total variance of each field first, and then performing the extraction of the first PC on the co-variance matrix of the combined fields (Wolter and Timlin, 1993). In order to keep the MEI comparable, all seasonal values are standardized with respect to each season and to the 1950-93 reference period.

IMPORTANT CHANGE: The MEI used to be updated every month during the first week of the following month based on near-real time marine ship and buoy observations (courtesy of Diane Stokes at NCEP). However, this product has been discontinued as of March 2011 (ICOADS-compatible 2-degree monthly statistics). Instead, the MEI is now being updated using ICOADS throughout its record. The main change from the previous MEI is the replacement of 'standard' trimming limits with 'enhanced' trimming limits for the period from 1994 through the current update. This leads to slightly higher MEI values for recent El Niño events (especially 1997-98 where the increase reaches up to 0.235 standard deviations), and slightly lower values for La Niña events (up to -.173 during 1995-96). The differences between old and new MEI are biggest in the 1990s when the fraction of time-delayed ship data that did not enter the real-time data bank was higher than in more recent years. Nevertheless, the linear correlation between old and new MEI for 1994 through 2010 is +0.998, confirming the robustness and stability of the MEI vis-a-vis input data changes. Caution should be exercised when interpreting the MEI on a month-to-month basis, since the MEI has been developed mainly for research purposes. Negative values of the MEI represent the cold ENSO phase, a.k.a.La Niña, while positive MEI values represent the warm ENSO phase (El Niño).

IMPORTANT ADDITION: For those interested in MEI values before 1950, a 'sister' website has now been created that presents a simplified MEI.ext index that extends the MEI record back to 1871, based on Hadley Centre sea-level pressure and sea surface temperatures, but combined in a similar fashion as the current MEI. Our MEI.ext paper that looks at the full 135 year ENSO record between 1871 and 2005 is available online at the International Journal of Climatology (Wolter and Timlin, 2011).

Historic La Niña events since 1950

Click to enlarge

How does the 2010-12 La Niña event compare against the six previous biggest La Niña events since 1949? This figure includes only strong events (with at least three bimonthly rankings in the top six), after replacing the slightly weaker 2007-09 event with 2010-12 (rankings are listed here). La Niña events have lasted up to and over three years since 1949, in fact, they do tend to last longer on average than El Niño events. The longest two events included here lasted through most of 1954-56 and 1973-75. The longest event NOT included here occurred in 1999-2001 which reached the 'strong' threshold (top six rankings) just once. Click on the "Discussion" button below to find a comparison of recent El Niño conditions with historic strong El Niño events.

Historic El Niño events since 1950

Click to enlarge

How does the 2009-10 El Niño event compare against the seven previous biggest El Niño events since 1950? This figure includes only strong events (with at least three bimonthly rankings in the top six), with the exception of the 2009-10 event that reached the top six ranking twice. Compared to the previous version of this figure, 1997-98 now reaches very similar peak values to the 1982-83 event, just above the +3.0 sigma threshold. Click on the "Discussion" button below to find a comparison of recent El Niño conditions with the same seven historic El Niño events. Once the 2015-16 event is over, the comparison figure with 2015-16 will replace the current one with 2009-10.

MEI loading maps for the latest season

Click to enlarge

The six loading fields show the correlations between the local anomalies of each field and the MEI time series. Land areas as well as the Atlantic are excluded and flagged in green, while typically noisy regions with no coherent structures and/or lack of data are shown in grey. Each field is denoted by a single capitalized letter and the explained variance for the same field in the Australian corner.

The sea level pressure (P) loadings show the familiar signature of the Southern Oscillation: low pressure anomalies in the west and high pressure anomalies in the east correspond to negative MEI values, or La Niña-like conditions. Consistent with P, U has positive loadings centered on the dateline, corresponding to easterly anomalies along the Equator from Indonesia and the Phillippines to 140W. In contrast, significant negative loadings cover the easternmost Pacific off the Central American coast as well as western Indonesia, denoting westerly anomalies during La Niña at this time of year. The meridional wind field (V) features scattered negative loadings north of the Equator across the Pacific basin, flagging the northward shift of the ITCZ so common during La Niña-like conditions, juxtaposed with large positive loadings northeast of Australia (northerly anomalies during La Niña).

Both sea (S) and air (A) surface temperature fields exhibit the typical ENSO signature of a wedge of positive loadings stretching from the Central and South American coast to the dateline, or cold anomalies during a La Niña event. Substantial negative loadings north and east of Australia contribute significantly to the overall temperature (S, A) pattern. At the same time, total cloudiness (C) tends to be decreased near the dateline over the west-central tropical Pacific and on the northeastern flank of the South Pacific Convergence Zone (SPCZ), sandwiched in between decreased cloudiness over Indonesia and the eastern-most equatorial Pacific.

The MEI now stands for over 30% of the explained variance of all six fields in the tropical Pacific from 30N to 30S, having regained about 13% since May-June. Nineteen years ago, right after the MEI was introduced to the internet, the explained variance for September-October was about 2% higher. The loading patterns shown here resemble the seasonal composite anomaly fields of Year 0 in Rasmusson and Carpenter (1982).

MEI anomaly maps for the latest season

Click to enlarge

With the MEI dropping deeper into the negative range, there are an unusual number of key anomalies in the MEI component fields that exceed or equal one standard deviation, or one sigma (compare to loadings figure), flagging both El Niño and (mostly) La Niña.

Significant positive anomalies (coinciding with high negative loadings) indicate high sea level pressure (P) over the eastern equatorial Pacific, increased westerly wind anomalies (U) over western Indonesia, southerly wind anomalies (V) across the northern subtropical Pacific, high sea surface temperatures (S) and air temperatures (A) north and east of Australia, as well as increased cloudiness (C) over Indonesia. Significant negative anomalies (coinciding with high positive loadings) flag lowered sea level pressure (P) over the Maritime Continent, increased easterlies (U) along the Equator near the dateline, and below-normal SST south of Hawai'i. All of these anomalies indicate emerging La Niña conditions.

On the other hand, a small region of enhanced easterlies (U) over southern Indonesia, lingering above-normal sea surface (S) and air temperatures (A) over the eastern subtropical Pacific, as well as reduced cloudiness (C) near Galapagos are more typical for El Niño than for La Niña.

Go to the discussion below for more information on the current situation.

If you prefer to look at anomaly maps without the clustering filter (which is most limiting for the cloudiness field), check out the climate products in our map room.

Discussion and comparison of recent conditions with historic El Niño conditions

Click to enlarge

In the context of strong El Niño conditions from April-May 2015 through April-May 2016, this section features a comparison figure with the classic set of strong El Niño events during the MEI period of record.

Compared to last month, the updated (September-October) MEI has dropped further to -.38 (down by -.28 in just one month), which translates into low ENSO-neutral rankings. The nine-month run in the Top-3 from May-June 2015 through January-February 2016 is tied with 1982-83 for its duration, while 1997-98 kept this level going for a full 12 months. No other El Niño since 1950 even exceeded three months at that level. The August-September 2015 MEI of +2.53 represents the peak of the 2015-16 event, and was exceeded only during the 1982-83 and 1997-98 events. The overall evolution of the 2015-16 El Niño was most similar to 1997-98, as monitored by the MEI, except that the latter had already transitioned to weak La Niña conditions by September 1998.

Looking at the nearest 12 rankings (+6/-6) in this season AND requiring that the MEI has dropped over the previous month as well as previous three months gives us six 'analogue' cases. Four of them ('61, '67, '95, and '05) ended up with at least weak La Niña during the subsequent winter, while two of them remained ENSO-neutral for much of that season ('00, '01). In other words, the odds for (weak) La Niña in the next few months have increased compared to earlier this year.

Positive SST anomalies cover much of the off-equatorial tropical Pacific, but cold anomalies have increased in coverage and are present right along the Equator from South America to the dateline, as seen in the latest weekly SST map. Thus, the cold anomalies look much more widespread than just the Niño 3.4 SST region.

For an alternate interpretation of the current situation, I recommend reading the NOAA ENSO Advisory which represents the official and most recent Climate Prediction Center opinion on this subject. In its latest update (13 October 2016), ENSO-neutral conditions were diagnosed, but a La Niña watch was reintroduced with odds for La Niña of about 70% during the remainder of 2016. Given my discussion above, I concur.

There are a number of ENSO indices that are kept up-to-date on the web. Several of these are tracked at the NCEP website that is usually updated around the same time as the MEI, in time for this go-around in the OISSTv2 version. Starting in May 2016, Niño region 3.4 SST anomalies dropped below +0.5C for the first time since late 2014. They reached -0.5C in July, and have now reached -0.7C in October. For comparison, Niño 3 SST has been a bit more reluctant to reach negative anomalies of this size, with October 2016 coming in at -0.4C.

For extended Tahiti-Darwin SOI data back to 1876, and timely monthly updates, check the Australian Bureau of Meteorology website. This index has often been out of sync with other ENSO indices in the last decade, including a jump to +10 (+1 sigma) in April 2010 that was ahead of any other ENSO index in announcing La Niña conditions. In 2016, its value varied from -22 in April to +14 in September, but fell again to -4 in October. As mentioned before, it is often out of sync with other ENSO indices.

The next update for the MEI will hopefully be about the same time in December as this month (the 8th), still delayed due to lingering issues with the update schedule of ICOADS data. El Niño conditions in the MEI sense have been replaced by ENSO-neutral conditions for four months running. Meanwhile, the PDO weakened dramatically from +2.6 in April to +0.5 in August and September, one of the biggest drops on record. Daily updates of the ENSO status can be found at the TAO/TRITON website, confirming the recent increase in coverage of eqautorial cold anomalies, along with more noticable easterly wind anomalies west of the dateline.

MEI data access and publications

You can find the numerical values of the MEI timeseries under this link, and historic ranks under this related link.

If you have trouble getting the data, please contact me under (Klaus.Wolter@noaa.gov)

You are welcome to use any of the figures or data from the MEI websites, but proper acknowledgment would be appreciated. Please refer to the (Wolter and Timlin, 1993, 1998) papers below (available online as pdf files), and/or this webpage.

In order to access and compare the MEI.ext against the MEI, go here.

Publications

  • Rasmusson, E.G., and T.H. Carpenter, 1982: Variations in tropical sea surface temperature and surface wind fields associated with the Southern Oscillation/El Niño. Mon. Wea. Rev., 110, 354-384. Available from the AMS.
  • Wolter, K., 1987: The Southern Oscillation in surface circulation and climate over the tropical Atlantic, Eastern Pacific, and Indian Oceans as captured by cluster analysis. J. Climate Appl. Meteor., 26, 540-558. Available from the AMS.
  • Wolter, K., and M.S. Timlin, 1993: Monitoring ENSO in COADS with a seasonally adjusted principal component index. Proc. of the 17th Climate Diagnostics Workshop, Norman, OK, NOAA/NMC/CAC, NSSL, Oklahoma Clim. Survey, CIMMS and the School of Meteor., Univ. of Oklahoma, 52-57. Download PDF.
  • Wolter, K., and M. S. Timlin, 1998: Measuring the strength of ENSO events - how does 1997/98 rank? Weather, 53, 315-324. Download PDF.
  • Wolter, K., and M. S. Timlin, 2011: El Niño/Southern Oscillation behaviour since 1871 as diagnosed in an extended multivariate ENSO index (MEI.ext). Intl. J. Climatology, 31, 14pp., 1074-1087. Available from Wiley Online Library.

Questions about the MEI and its interpretation should be addressed to:
(Klaus.Wolter@noaa.gov), (303) 497-6340.